The Hidden Data Problem Behind Institutional Crypto w/ Kiet Tran

Transcript

Please welcome Kit Trann, CEO of Luca, in conversation with Stablecon's [music] Dante Reminic for a special stable pulse recording sponsored by Luca. Hey everyone, welcome to Stable Pulse. We have a really fun conversation here today. We're going to dive into data, how data interweavves itself into pretty much anything that digital assets touch. It's going to be very difficult for me to keep this to an aotted time kit because I feel like we can go on forever ever and ever. There's a lot of a lot of rabbit holes to go down. But let's just kick things off with an introduction.

You have a very very impressive background. You've already done some very impressive things at Luca. Brag about yourself. >> Uh well, bragging is not something I like to do, but I'll I'll try my best. Uh >> look, so I you know, I I started my career at Black Rockck Aladdin. Um you know one of the things I tell people is sometimes you better be lucky than to be good. In my instance I was lucky right place right time. Um and then in early05 I joined a startup called market which became IS market right which was acquired by S&P global for $44 billion.

But during my journey there I spent a lot of time around market infrastructure market structure especially around the OTC marketplace. And one of the things we talked about was you know I was the team that actually built the ability to short mortgages. So you read the big short I dealt with all those cast of characters on a daily basis and I I do think you know when we talk about digital asset in general I think it's really part of market transformation I think overall global markets we started with analog to I would say electronic in the 70s with NASDAQ TTCC and I think the industry's gone digital the 2000s with web APIs mobile etc and now my view is we're in the next phase which I called

programmable markets. Imagine having assets and investments that have programmable logic but also leveraging agents to transact and trade right which we've seen with Robin Hood and other platforms. So I think that's the phase that we're in right now which is quite exciting. And what's exciting about Luca is you think about Luca at a macro level. Look, Luca is a global data and software infrastructure provider for digital assets. We help institutions understand value reconcile govern and report digital asset with the same rigor that you need for traditional finance. And the one thing people don't talk about is if I buy a stable coin or

buy a digital asset, well, you still need to reconcile it back to your ERP system, your treasury system, your investment book record system, your accountable book record system. that doesn't go away, right? And and that's the superpower that Luca provides. And we do this for a lot of infrastructure provider, fund administrators, exchanges, asset managers, banks. And so for us, like where the plumbers in suits that people don't talk about, right? But that's what we do. >> But everyone needs. >> Exactly. Exactly. I love that. And let's talk about what that data actually looks like. To your point, you know, we've moved from an analog world to a digital world. Now this is you know the the second coming of of you know progress of financialization online. How is the data

that people need to to know when interacting with these digital assets? How has that changed since you know the first wave of innovation that's come through Wall Street? I think you know there's the decentralization aspect which on the opposite side mean this lack of standard. Yeah. something as simple as a Bitcoin, there's 30 plus tickers for Bitcoin, right? And so when you think about like if I want to buy or sell a bond, if I want to buy or sell a stock, there's only certain amount of exchange venues I can do that with. But if I want to buy crypto, there's a lot of C5, D5 exchange, right?

We're connect to hundreds of venues. And so how do you normalize all of that, right? That's just the offchain stuff. And then you also then have the all the onchain stuff. And when you think about the onchain stuff, let's say Salana, if I want to pull every single transaction for Salana down to the Genesis level, that's before indexing. >> Mhm. >> That's 4 to 500 terabytes worth of data. Massive amounts. And so what Bluca has come to be in and what we realize is because of our ability to understand and normalize both off and onchain data, it helps us give institutions the ability

to understand what asset are you holding, right? What's backing this asset? How do you reconcile your position? How do you understand the counterparts you're dealing with? And so to me it's a big data challenge, right? And and that's something that we've focused a lot of our energy on. >> Yeah. I think that there's this really interesting sort of dichotomy that happens with onchain digital assets because on one hand you have the transparency of a blockchain, right? Everything is is recorded on that chain. And on the other hand, it's almost so much data that it's hard to understand the actual actions behind that data. How do you go about collecting terabytes and terabytes and terabytes of information

and producing something that's workable, that's usable, that's actionable for your clients? Yeah, I I think a lot of it is our ability to triangulate and normalize and we're able to do that because a lot of the global fund administrators. So, you know, you think about like hundreds of hedge funds or asset managers that buy and sell these things, they need a fund administrator. Well, we're the ones that service the fund administrator. So, because of that, you know, we understand all the assets. We understand the reconciliation of positions. So that give us an edge and because we also run our own blockchain

nodes and we also do our own indexing when you pull this together it gives you a view that no other can have. Right? So you think about for example we talking about hey everything's on the blockchain it is but let's say I have a Solana wallet and on this wallet there's six assets with this smart contract. Like well what asset is that? How do you link back? Is this tokenized Nvidia? >> Yeah. Is this a bitcoin? Like what is this? >> Yeah, you have to follow the trail. >> You have to follow the trail. So, so we provide the linking and mapping of this trail, >> right? To make everyone's lives a lot easier. >> And there's one thing that you mentioned that I want to circle back on, which is the connection between the onchain system and the off-chain system, right?

Anyone who's working with digital assets has to have both. There is no fully onchain solution. There's no fully offchain solution. Tell me about the challenges that most people face when dealing with mirroring the onchain and off-chain solution and how you guys go about solving that. >> I think a lot of it is um it's really about reconciliation and understanding one realizing that you know issuing a smart contract and tokenizing is the first of many step. You think about a bank well you got to connect back to the 30 systems the bank's sitting on. Yeah.

Right. And so it's really about building that connectivity on what it means onchain, what it means offchain. Let's use Nvidia as an example, right? Nvidia is a stock. Well, Nvidia has a stock has a stock taker. It trades on these exchanges. It has an icing, has a QIP, it has a CEDO. And what we do is we then map that to then all the tokenized versions of Nvidia, right? That's onchain. We mapped that to all the prediction markets that are related to Nvidia. We map all the perpetual futures in Nvidia. So then give you a 360 view. So when you do the reconciliation, you now have an apples to apples linkage.

But then there's more to that. And what I mean by that is in the traditional world of how we deal with like a stock has a corporate actions etc. on the blockchain location matters and so I I do think it's the decision of how we build our data model where we will have a an identifier for an asset. Let's look Bitcoin, we have a global identifier, but Bitcoin on Ethereum, we also have an identifier. So, we have an identifier at the asset location. So, then that gives us this deep linkage back to all the breadcrumbs. And so, anyone that wants to reconcile on and off chain, they can use all these breadcrumbs to reconcile. And that's going to be important, right? Because

look, your accounting book record system, your investment book record system, your it's not going away. You're not changing that overnight. >> Yeah. >> Right. In fact, it'll probably get more complicated as more assets come on chain, more issuance, more more money moving on a blockchain on blockchains. I imagine this there's just compounding complexity to what you're doing. So, the the way I put this is um and you're too young, maybe I'm too old. You think about like the early internet days, right? In the early internet days, you know, people have an a store and then they sell online, but they have an online system, inventory system, they have a store inventory system. they don't talk to each other >> which means they don't have any system.

So in today you have your C5, your D5, your Trafy and they don't reconcile you don't have a ledger, right? So that needs to be reconciled and that's really a big part of where Lucas sees our value in the future. So let's talk about this and honestly there is probably a more tactful way to ask you this question. But who cares, right? like when it comes to to digital assets and data, what are the negative externalities of not working with someone like Luca? Like if you know I have a hedge fund or I have a financial institution and I want to go trade, what can happen if I don't maniacally focus on this level of reconciliation and building out this

comprehensive data layer. >> I mean, you don't understand your risk [clears throat] profile, you don't understand your positions, how are you reporting back to your regulators, right? And one of the things we know is there's more and more regulatory clarity which means you need to report this back to all your global regulators right whether it's cough 109A ma there's all these regulation you need to align back to but also just smart way of running your own business. >> Yes. >> Like why would you run a business and not be able to reconcile all all your inventories? Absolutely right. And so to me any institutions and look this this is really the the premise of Luca's founding right

>> um Luca was originally known as a Bitcoin tax calculator. We founded in 2014. >> Interesting. >> And then we realized that institutions will step into this space, right? And that that's how my focus is really more around crypto, right? And so what we did was we say, well, institutions were to participate in this asset class and this marketplace. They would need to be familiar with how they operate in other markets. You need a reference data layer. You need a corporate access layer. You need a pricing evaluation layer that's actually IFS gap compliant. That's key. So what's interesting is all these digital asset startups that are going public, the

first thing that ought to have them do is call Luca because we're still the only after gap compliant valuation service out there. Right? >> That's amazing. >> And then how do you build your ledger system on tracking your investment book record, your account broker record. So we built the full stack with the view that institutions will come in and what we realize is the opportunity is not in crypto because crypto is the first asset class tested on the blockchain and tested on digital asset. We think the future rail is digital asset >> and and so >> can you tell me more about that?

>> Why you say the future rail is digital assets themselves? I I think look I I think the reality is the the biggest unlock is really around 24/7 collateral unlock the efficiencies that it brings right but let's not confuse that as an asset class it's not an asset class because you will have tokenized versions of fixed income tokenized versions of equities which you see the NASDAQ TTCC n running through you know as my point is you know market's going to be programmable but not every asset class will move the same speed of direction but it's quite timely that we're stable coin today I think

stable coin is what's driving a lot of these innovation because what is a stable coin at a very macro level is an atomically settled liability you need to assets back in it which is short-term treasuries repos money mark funds that's still T+1 t+2 so what you got to tokenize that right so I think these drivers of really showcasing the value of digital asset as a rail. >> I I really like that mental model and something you and I were talking about backstage of stable coins as a connection layer between disparate systems, right? You might have an RWA issuer over here, an exchange over there, a different RWA issuer over on the other end. And the way to connect

all of these disparate systems is is with stable coins. So, I'm a I'm a really really big fan of that mental model. We've talked a lot about using data retroactively for reconciliation. Are people also using the data that you're collecting for future actions to inform, you know, what moves they should be taking in the future, not just reconciling what's happened in the past. >> It varies. Look, so we we also power a lot of uh hedge funds and obes. I think part of the challenge is we do great work behind the scenes. We don't go around bragging because that's not, you know, what we do best at, but maybe we need to do a better job bragging. Um, so a lot of objects do use us, right?

And because we have a diverse type of data set that allows them to triangulate, right? Imagine understanding what's happening in prediction market, perpetual market or tokenize a real world assets. It gives them an edge that they otherwise wouldn't have. >> Interesting. I imagine that having all this data is not only a performance advantage, but there's also a risk and regulatory advantage. We've talked a little bit off the stage about market structure, what's happening in the United States, what's happening around the world, just in terms of, you know, forcing financial institutions to comply with various different laws around, you know, you're able to use this issuer, not this issuer. You know, this is what a a good digital asset practice looks like. How have you guys needed to adapt with, you know, the market structure

that's coming up around the world in terms of the type of data that you collect or the type of data that your clients want? >> Yeah, I think a big part of that is, you know, one of the things that our team has harnessed over the last year, so since taking over Luca, you know, my view is if you're not an AI company, you're not a company. Yes. >> And so we've invested heavily in our AI infrastructure and really give us that flexibility to operate. And so behind the scenes on what runs everything is is a platform called Luca intelligence. So think of it as you know cla or open codeex but for capital markets workflow and our team runs on it.

>> You know we have our okadomo agent that goes and and really uh spin up nodes. uh we have a recon agent and so a lot of these tech advancements has allowed us to really accelerate what we do and if you think about it's not just about the AI it's about the data power in the AI so because we're sitting in so much data it give us that tremendous advantage can we go down this agent rabbit hole for a little bit I want to get a better understanding of how you guys are using agents to collect decipher all this data But also on the other side, right, are your are your clients using agents? Are those agents being informed by the data

that that you're collecting? Walk me through all of it. >> Well, what's interesting is we we haven't exposed this to our clients yet. >> Oh, >> but uh but we've demoed to them and they want it, >> of course. >> Right. >> Um look, for for us, it's really about how do you harness deterministic workflow and probabilistic workflow. The ELM itself is a probabistic workflow, right? Give me the next word. Give me the next context, right? But when you're reconciling, you know, you can't run a financial model and say, well, the answer could be this or that or that, >> right? It needs to be deterministic.

>> And for an AI workflow to be functional, you need what I deem as the ASUS. What does that mean? It needs to be accurate. It needs to be consistent. It needs to be explainable. And it needs to be secure. And our view is there's no better way to prove it than to run this ourselves with the workflow we run on. And we're now at a stage where we're so confident we're in the process of rolling this out to our client base. >> Amazing. >> And that's the that's the power really about leveraging and harnessing the technology with the deep insights the team bring.

>> Yeah. And one of the things you know people don't realize is the way I use LM versus the way someone else you may be different because I have different context and I do think the next phase of data it's about context on how to do something right how do >> you see your job as my job is to get the best context for my clients >> that's exactly it >> I love that >> right we're context engineers that's what we are so >> it's a really interesting way of of looking at what what Luca is does it stop at context or do you guys take that context and and turn that into insights or do you just give your clients the tools to do that themselves?

>> It's it's both. Look, I I I I think the the future is going to be headless in a way, right? So So all solutions we have all the MCPS MCPS is available on anthropic marketplace, right? >> The Luca MCPS are >> Yeah. Yeah. >> Amazing. Um again we need to do a better job marketing that but we do have that right. Um, but we've also harnessed our own LM and run our own stack internally. And one of the things we realized is that the more we run our workflow on, the more we're giving it context, right?

And so we've been hosting our own open weight model and we've been building what I call the Luca brain. And the more we inject context into the brain, what it does is really put guard rails to make sure it doesn't go haywire, right? And to make sure that the output is deterministic even though the workflow combines deterministic tooling with probabistic solutions in the mix. I imagine though that the complexity of this context is something that you have seen grown tremendously. again, always more chains, always more assets on chain, always more issuers coming and issuing 17 versions of of SpaceX stock.

How has that challenged your team to build out these these contextual models differently as as just the snowball of complexity? >> Well, well, to add to that, right, so it's funny. One of my good friends, you know, they launched a stable coin uh last year. It's a same day 10 stable coins came out of course trying to pretend to be us, right? Of course, >> that's the other level of complexity you need to solve. I I think the advantage we have is because of the reference data >> and and a lot of the work we put in place. It gives us that edge. So, we have a massive deterministic set of tooling and database and marrying that with these probabilistic tools and hosting open weight models allow us to triangulate that we otherwise couldn't.

Right? So you think about tracking and tracing and being able to have a deep understanding on the 1.2 billion wallets that we track and trace. Marrying that with these AI tooling will give you superpower you otherwise didn't have. And so for us it's really about embracing this new tooling and going all in on it. I like that. So that's I imagine what most of your team spends its time on is you know you advertise yourselves as hey we're we're a tool for people who are working with digital assets but on the back end you guys are entirely focused on how to optimize with agents and everything like that.

>> Exactly. >> Do you see your clients being as aggressive on the AI front and on the agent front as you guys are? >> It I think it varies on the types of accounts. Look, I I think everyone went from token maxing >> Yeah. >> to now value maxing. >> As it should be. >> As it should be. Yeah. Look, the the issue is is why would you give your data, give your context to Frontier models and pay them at the same time. >> There's nothing worse than the Toys R Us moment, right? When when Toys R Us taught Amazon how to be in a toy business and Amazon, well, thank you very much. We're now in the toy business.

>> Yeah. And so the the next phase is really around AI sovereignty. But the reality is I I think part of the challenge we seeing look for very large institutions and I may get in trouble for saying this but you know the executive level to the person that knows what they're doing there's like probably six to eight levels in between of middle management >> and so every company's going through a different journey. I've seen some firms that shut off all AI access cuz they're freaked out to other firms say you know what let's play with it let's figure out and so there's no one size fit all but the reality is

in today's world with all these model it's like keeping up with the Joneses >> so Astra just came out right Astra the open eye version six whatever the case is >> um you know you have a new version of an LM a model everywhere with all all these open model. So how do you know which one to use? Like it's intimidating. It's confusing which is similar than all this blockchain, right? All these different chains, all these different access. >> There's two incredibly complex systems that are coming together. >> That's exactly it. So our view is we just help harness this on behalf of our clients. So they don't need to worry about >> which to use, which system to harness.

Look, put it this way. If I need to know who's the first president of the United States, I don't need fable. The dumbest model will in theory know the answer. >> Yes. >> So, how do you harness that orchestration across workflow? And our view is, you know, we're we're using this for our own cooking deeply internally so that we'll offer this out to our customers. And and that's really been the advantage is, you know, everyone is an AI expert across Luca, right, from that perspective. I love that. And again, backstage, you and I were talking about how it's still the very very early days of Luca. You've gone out, you've built this incredible system that marries the

complexity of digital assets and the complexity of AI. You're actively delivering value to some of the world's largest financial institutions, but it's still the early days. What's next? Look I I think financial institutions are getting the joke. You think about this whole stable con today. How many banks you see here, right? And so if you're a financial institutions and you don't treat digital asset or stable coin seriously then you're sleeping under the rocks and I think that leads to a lot of opportunities for Luke and what we built right we've been building to waiting for this moment and the moment is here so it's quite exciting for us you know the

team is super busy um and I I do think this tailwind continue to be an opportunity for us >> yeah I mean to your point you and I have spent the last 48 eight hours straight talking to people who are building out stable coin programs or in one way or another, but whether they are a hedge fund that wants to ar between stable coins, whether they're a a B2B stablecoin orchestration company, I think that where the ven diagram converges is that every single one of those companies needs data. Like you you cannot unequivocally run any type of digital asset program without having the data to back that up. Have you found that the type of data that is needed across these variety of different

clients changes or does everyone need roughly the same type of data? >> So it's not just about data but I'll say data infrastructure. >> Okay. >> Right. Because you need to map back to all the onchain data >> and it's a massive amount of data. Right? Luca is sitting on 10 plus pabytes worth of data. So it's a massive data set, but it's also a massive exercise of mapping all the breadcrumbs. And so I think that's the advantage that Luca brings to the table is all the breadcrumbs that we map to give you that end to end linkage. And so think of us as not just data, but data infrastructure.

>> Yeah. And I I do think that's something that everyone's going to need if you want to be able to reconcile back and report back to your investors and regulators, you know, your positions, your risk, exposure, etc. That's going to be important. >> Yeah. I mean, I talk about we we talk about the stable coin stack and often times that stack is made out of various different infrastructure components. You have, you know, your wallets over here, you have your ramps over here, you have a a custody provider, an issuer, whatever that might be. Do you find that people are coming to the the data layer in the stack at the right time or is this often something that they realize you know after the fact uh oh you know now that I need to do all this reporting I need to go after the data the right

data >> yeah and in in a lot of case what they end up doing because they didn't set up right they throw a lot of bodies at it right things ends up being super manual >> yes >> and I do think that's the advantage we provide is you leverage Luca >> you no longer need that many people in that process you got to put in place right on this reconciliation nightmare you got to run. >> I I always find it funny when you know the the some of the most advanced financial organizations in the world, people who are working with agents, people who are working on the frontier of of trading or whatever it might be, still have a team of 10 working off of a shared spreadsheet to to reconcile all their team.

>> I was funny, I was talking with someone earlier today. It's ridiculous, but it's all over the place. >> But it's true. >> It's true. >> It's true. It's true. You know, I I see people using the most advanced technology, still running off of an Excel sheet and still doing OTC DR trades by by texting their banker and whatever. >> That's right. That's right. Signal, WhatsApp. [laughter] >> Exactly. It's it's the it's the dark duopoly of of you know, innovation in this space is that there are people who push and push and push the frontier, but also still still operate on, you know, really old humanentric uh actions, I would say. So hopefully you can do a lot to to eradicate that and just say, "Hey, run the most advanced system in the world and we'll handle the rest."

>> Exactly. >> Exactly. >> I love it. So there's going to be a lot of people who are listening to this who are building out these programs, who are at financial institutions, who are at B2B payment companies, whatever it might be. How should they get a hold of your team if they want to learn more? What do they do? What's the next step for everyone here? >> Yeah, look, look at that tech. We also have a LinkedIn page. uh definitely hit up our team. I I I do think once you get to know us as a team, once you get to see our platform, you're going to find out you've been missing out. I can confirm that their team is great. Uh and I also the one thing that I really like about your team is that you recognize that every data problem is different and

so you customize a lot of the work to the individual clients and I think that that's something that's really really special in the space. But Kit, I really appreciate you coming up here. Like I said, you and I could do this for three more hours, so we're going to have to have you back and, you know, dive into the nuances of market structure and and home mortgages and everything like that. But that's for another time. I appreciate you coming up and uh we'll see you soon. Thank you for having me. Have a good one.